Electrical Power and Energy Systems

Electrical Power and Energy Systems 44 (2013) 880 888 Contents lists available at SciVerse ScienceDirect Electrical Power and Energy Systems journal homepage: www.elsevier.com/locate/ijepes Profit based unit commitment and economic dispatch of IPPs with new technique H. Tehzeeb-ul-Hassan a, Ayaz Ahmad b, a Department of Electrical Engineering, University of Engineering and Technology, Lahore, Pakistan b National Transmission and Despatch Company Limited, Muzaffargarh, Pakistan article info abstract Article history: Received 7 April 2012 Received in revised form 21 July 2012 Accepted 19 August 2012 Available online 26 September 2012 Keywords: Unit commitment Economic dispatch Profit based unit commitment Competitive power market Each generation company may have number of generating units of different fuel consumption characteristics, some generating units consume more fuel as compared to other units this directly effects the production cost and profit of the company. Production cost and profit of the company is also affected by unit commitment and economic dispatch. Each and every power generation company wants to maximize/ increase profit, same is the case for independent power producers (IPPs). Profit can be maximized by changing the unit commitment and economic dispatch strategy. Previously it was achieved in such a way that production cost goes to minimum level. But as the competition in power market is going to increase day by day IPPs trend of UC solution is toward achieving maximum profit. Previously achieved solution by LR PBUC is slow and may face convergence problem. In this paper, we will see the way how a GENCO or IPP can earn more by UC on the base of profit, with minimum computational time and always with some final solution. Hamiltonian method has been used for ED. To demonstrate the effectiveness of the PBUC achieved and Hamiltonian economic dispatch, it will be tested on two test cases. Profit and computational time comparison of proposed technique with already available/techniques for evaluation of performance are also presented. Ó 2012 Elsevier Ltd. All rights reserved. 1. Introduction With the advent of restructured system it is possible for power generation companies and independent power producers (IPPs) to consider such a scheme/schedule in which they supply the amount of power that is near to the predicted load demand and spinning reserve [1]. The objective of the generation companies and IPPs is to generate and sale the energy with maximum profit to survive in the competitive environment [2,3]. This leads us to develop and implement those techniques of committing the units that are based on maximizing the profit instead of minimizing the production cost as the case was in previous years [4]. This technique is known as profit based unit commitment (PBUC). It increases the profit of company; as a result Power Company can compete in a better way. 2. Profit based unit commitment Corresponding author. Tel.: +92 312 4971533/3334971533/662512220; fax: +92 4299250202. E-mail addresses: tehzibulhasan@gmail.com (H. Tehzeeb-ul-Hassan), hello_ayaz @yahoo.com (A. Ahmad). With the idea of PBUC, unit commitment problem (UCP) defined in a new way and with modified constraints because now our target is to maximize profit instead of minimizing production cost. Different techniques were proposed by researchers for PBUC. A hybrid method of Lagrange Relaxation (LR) and Evolutionary Programming (EP) has been used by Pathom Attaviriyanupap for profit based unit commitment in competitive power market environment [5]. Yuan and Yuan implemented the Particle Swarm Optimization (PSO) technique for profit based UC under deregulated electricity market [6]. Chandram and Subrahmanyam introduced new approach with Muller method for PBUC with small execution time [7]. For generating units having nonlinear cost function Mori and Okawa proposed the new hybrid Meta-heuristic technique for profit based unit commitment [8]. Amudha and Christober Asir Rajan presented effect of reserve in PBUC using Worst Fit Algorithm [9]. According to the above discussion in PBUC, our objective is to maximize Power Company s profit. So: Max: profit ¼ Revenue Total production cost ð2:1þ Modified and unchanged constraints in achieving above objective for IPPs according to the new scenario can be defined as: (i) Demand constraints X N i¼1 A it B it 6 C t ; t ¼ 1; 2; 3; 4;...; T ð2:2þ (ii) constraints are X N i¼1 D it B it 6 SRt; t ¼ 1; 2; 3; 4;...; T ð2:3þ 0142-0615/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ijepes.2012.08.039

H. Tehzeeb-ul-Hassan, A. Ahmad / Electrical Power and Energy Systems 44 (2013) 880 888 881 Fig. 1. Flow chart of complete strategy. (iii) Power limit constraints 0 6 D 6 A i max A i min ; i ¼ 1; 2; 3; 4;...; N ð2:4þ D i þ A i 6 A i max ; i ¼ 1; 2; 3; 4;...; N ð2:5þ (iv) Minimum down times constraints B it ¼ 0 for X t 1 t T down¼1 ð1 B it Þ < T idown (v) Minimum up time constraints ð2:6þ (vi) Start up cost constraints SUC ¼ constant if B i ; ðt 1Þ ¼0 and B it ¼ 1 ð2:8þ (vii) Shut down cost constraints SDC ¼ constant if B i ; ðt 1Þ ¼1 and B it ¼ 0 ð2:9þ There are many types of payments in power market. We have only considered the Payment for Power Delivered. In this case payment will be made when power will be used actually; revenue and cost can be calculated as [1]: B it ¼ 1 for Xt 1 t T up B it < T up ð2:7þ Revenue ¼ XN X T ða it SPtÞB it þ XN X T i¼1 t¼1 i¼1 t¼1 r RPt D it B it ð2:10þ

882 H. Tehzeeb-ul-Hassan, A. Ahmad / Electrical Power and Energy Systems 44 (2013) 880 888 Fig. 2. Flow chart of economic dispatch strategy. Cost ¼ð1 rþ XN X T i¼1 t¼1 FðA it ÞB it þ r XN X T i¼1 t¼1 FðA it þ D it ÞB it þ SUC B it ð2:11þ where A it is the power output of generator i at t hour, B it is the on/ off status of generator i at t hour, C t is the total demand at t hour, D it is the reserve generation of generator i at t hour, SRt is the spinning reserve at t hour, SPt is forecasted spot price at t hour, RPt is forecasted reserve price at t hour, SUC is startup cost, k t and l t are Lagrange multipliers, r is the probability of calling the reserve. LR looks to be the most appropriate method for the solution of large scale PBUC problems but it is hard to always get the final solution and with reasonable computational time, because of convergence and large number of iterations [10]. Lagrange function is formed by taking/assuming k t and l t as Lagrange multipliers to the demand and reserve constraints: LðA; D; k; lþ ¼cost revenue XT t¼1 ktðc t XN A it B it Þ i¼1 XT ltðsrt XN D it B it Þ t¼1 i¼1 ð2:12þ Fig. 3. Computational/execution time comparison of different techniques with proposed technique.

H. Tehzeeb-ul-Hassan, A. Ahmad / Electrical Power and Energy Systems 44 (2013) 880 888 883 Fig. 4. Profit comparison of proposed technique with other techniques. Table 1 Forecasted spot and reserve price of each interval (hour) for three unit 12 h system [5]. Hour Forecasted spot-price ($/MW h) Forecasted reserve-price ($/MW h) Hour Forecasted spot-price ($/MW h) Forecasted reserve-price ($/MW h) 1 10.55 31.65 7 11.30 33.9 2 10.35 31.05 8 10.65 31.95 3 9 27 9 10.35 31.05 4 9.45 28.35 10 11.2 33.6 5 10 30 11 10.75 32.25 6 11.25 33.75 12 10.6 31.8 Table 2 Forecasted demand and reserve of each interval for three unit 12 h system (hour) [5]. Hour Forecasted demand (MW) Forecasted reserve (MW) Hour Forecasted demand (MW) Forecasted reserve (MW) 1 170 20 7 1100 100 2 250 25 8 800 80 3 400 40 9 650 65 4 520 55 10 330 35 5 700 70 11 400 40 6 1050 95 12 550 55 Table 3 Units constraints of IEEE 3unit 12 h test system. Unit P max P min Start up cost ($) Min down time (h) Unit1 600 100 450 3 3 Unit2 400 100 400 3 3 Unit3 200 50 300 3 3 Min up time (h) Table 4 Fuel cost function parameters of IEEE 3unit 12 h system. Unit A B C Unit1 0.002 10 500 Unit2 0.0025 8 300 Unit3 0.005 6 100 Using the dual optimization technique coupling constraints are relaxed and reach the optimum by maximizing Lagrange function L with respect to the Lagrange multipliers k t and l t, while minimize with respect to the control variable A it, D it and B it, that is: q ¼ maxðk t ; l t Þ qðk t ; l r Þ¼min w:r:t B; A; DLðB; A; D; k; lþ ð2:13þ ð2:14þ Over all time periods by solving the minimum for each generating unit, we get minimum of Lagrange function. Accuracy/quality of result depends on the procedure that is used for updating the multipliers. In LR for unit commitment (UC), problem is divided into sub problems. By attaining the solution of each sub problem we reach to the actual or final solution of the problem. By just taking objective function information and initiating from different points we use stochastic optimization techniques like EP for searching the solution. A combined technique of LR and EP is used for solving the PBUC problem. It is done by dual optimization technique and where coupling constraints are relaxed/ignored. For enhancing the efficiency of LR, LR multipliers are updated by EP. Though it enhances the efficiency but still it may take large execution time and can face convergence problem [1].

884 H. Tehzeeb-ul-Hassan, A. Ahmad / Electrical Power and Energy Systems 44 (2013) 880 888 Table 5 Active power and reserve generation of Proposed PBUC and HED. Proposed technique of PBUC and HED Hour Power (MW) (MW) Profit ($) Hour Power(MW) (MW) Profit($) Unit 1 Unit 2 Unit 3 Unit 1 Unit 2 Unit 3 Unit 1 Unit 2 Unit 3 Unit 1 Unit 2 Unit 3 1 0 0 170 0 0 20 531.385 7 0 400 200 0 0 0 1380 2 0 0 200 0 0 0 570 8 0 400 200 0 0 0 990 3 0 0 200 0 0 0 300 9 0 400 200 0 0 0 810 4 0 0 200 0 0 0 390 10 0 130 200 0 35 0 818.101 5 0 0 200 0 0 0 500 11 0 200 200 0 40 0 804.63 6 0 400 200 0 0 0 950 12 0 350 200 0 50 0 929.231 Total 8973.3 Computational time (s) 0.06 Table 6 Generated and reserve power of traditional UC [1]. Traditional unit commitment Hour Power (MW) (MW) Profit ($) Hour Power (MW) (MW) Profit ($) Unit 1 Unit 2 Unit 3 Unit 1 Unit 2 Unit 3 Unit 1 Unit 2 Unit 3 Unit 1 Unit 2 Unit 3 1 0 100 70 0 0 20 126.5 7 500 400 200 100 0 0 1040.9 2 0 100 150 0 0 25 352.9 8 200 400 200 80 0 0 548.4 3 0 200 200 0 40 0 103.6 9 100 350 200 15 50 0 308.1 4 0 320 200 0 55 0 303.1 10 100 100 130 0 0 35 91.1 5 100 400 200 70 0 0 363.2 11 100 100 200 0 40 0 159.7 6 450 400 200 95 0 0 1017.8 12 100 250 200 0 55 0 359.9 Total 4048.8 Table 7 Active Power generation of Muller Method of PBUC [7]. Time (h) Muller Method for PBUC Unit Commitment Without Profit ($) Time (h) Muller Method for PBUC Unit commitment Without Power(MW) Power(MW) Unit 1 Unit 2 Unit 3 Unit 1 Unit 2 Unit 3 1 0 0 170 529 7 0 400 200 1380 2 0 0 200 570 8 0 400 200 990 3 0 0 200 300 9 0 400 200 810 4 0 0 200 390 10 0 130 200 813.75 5 0 400 200 200 11 0 200 200 800 6 0 400 200 1350 12 0 350 200 923.75 Total 9030.5 Computational time (s) 0.078 Profit ($) Table 8 Forecasted spot and reserve price of each interval (hour) for ten unit 24 h system [5]. Hour Forecasted spot price ($/MW h) Forecasted reserve price ($/MW h) Hour Forecasted spot price ($/MW h) Forecasted reserve price ($/MW h) 1 22.15 110.75 13 24.6 123 2 22 110 14 24.5 122.5 3 23.1 115.5 15 22.5 112.5 4 22.65 113.25 16 22.3 111.5 5 23.25 116.25 17 22.25 111.25 6 22.95 114.75 18 22.05 110.25 7 22.5 112.5 19 22.2 111 8 22.15 110.75 20 22.65 113.25 9 22.8 114 21 23.1 115.5 10 29.35 146.75 22 22.95 114.75 11 30.15 150.75 23 22.75 113.75 12 31.65 158.25 24 22.55 112.75 3. Economic dispatch After profit based unit commitment the second step is distribution of load on committed units. Load distribution on committed units must be optimum. The distribution of power demand on generating units to supply the load with minimum operational cost is known as Economic Dispatch (ED) [11]. In economic dispatch individual machine efficiency may not be maximum, but overall this

H. Tehzeeb-ul-Hassan, A. Ahmad / Electrical Power and Energy Systems 44 (2013) 880 888 885 Table 9 Forecasted demand and reserve of each interval (hour) for ten unit 24 h system [9]. Hour Forecasted demand (MW) Forecasted reserve (MW) Hour Forecasted demand (MW) Forecasted reserve (MW) 1 700 70 13 1400 140 2 750 75 14 1300 130 3 850 85 15 1200 120 4 950 95 16 1050 105 5 1000 100 17 1000 100 6 1100 110 18 1100 110 7 1150 115 19 1200 120 8 1200 120 20 1400 140 9 1300 130 21 1300 130 10 1400 140 22 1100 110 11 1450 145 23 900 90 12 1500 150 24 800 80 Table 10 Unit constraints data of 10 unit test system. Unit P max P min Start up cost ($) Min down time (h) Min up time (h) Unit1 455 150 4500 8 8 Unit2 455 150 5000 8 8 Unit3 130 20 550 5 5 Unit4 130 20 560 5 5 Unit5 162 25 900 6 6 Unit6 80 20 170 3 3 Unit7 85 25 260 3 3 Unit8 55 10 30 1 1 Unit9 55 10 30 1 1 Unit10 55 10 30 1 1 Table 11 Fuel cost function parameters data of 10 unit test system. Unit A B C Unit1 0.00048 16.19 1000 Unit2 0.00031 16.26 970 Unit3 0.00200 16.6 700 Unit4 0.00211 16.5 680 Unit5 0.00398 19.7 450 Unit6 0.00712 22.26 370 Unit7 0.00079 27.74 480 Unit8 0.00413 25.92 660 Unit9 0.00222 27.27 665 Unit10 0.00173 27.79 670 load distribution by economic dispatch will provide maximum/increased profit. Hamiltonian optimization is one of the important optimization techniques to solve dynamic, deterministic optimization problems. The Maximum Principle/Hamiltonian is also used for the solution of Economic Growth problems. The Hamiltonian principle has vide application in physics as well. We used the Hamiltonian technique for the solution of ED part. First of all problem is converted according to the general set up as discussed above. Then we have to follow seven steps for getting solution by this method. Each step is described below. 3.1. Step1 Develop the Hamiltonian function according to your specific problem. Add the felicity function to the Lagrange multiplier times the RHS of transition equation for generating the Hamiltonian function. H ¼ vðk; c; tþþlðtþf ðk; c; tþ ð3:1þ In our case, felicity function is the sum of cost function of all committed generating units and transition equation consist of load and reserve demand constraints. Where load demand constraint for ED is equality constraint while reserve demand constraint is inequality constraint. But according to our proposal, we call only the ED function when load is less than the total generation capacity of the committed units. 3.2. Step2 Differentiate H Hamiltonian function partially with respect to c control variable and put it equal to zero. Mathematically it can be written as: @H @v @c ¼ @c þ l @f @c ¼ 0 ð3:2þ In our case, p is the control variable. So, differentiate the function w.r.t. p. In third step, take derivative of Hamiltonian function with respect to state variable and we have already value of state variable from PBUC solution. In rest of the steps either we have to follow the previous steps or need time derivative. We evaluate ED for each time interval independently so, we need not to follow third and remaining steps. We get the solution by only following the first two steps. 4. Solution strategy For solution of PBUC and Hamiltonian ED, first step is to initialize I/P, O/P variables also assign the values to fixed parameters. Active, reserve power, on/off status, profit variables and k are initialized with zero value. Rest of the variables and fixed parameters are defined by input file. After initialization of the variables we calculate lambda for each generating unit where it is feasible to start. This value is calculated by the following formula: k i ¼ cost of unit at P max;i P max;i þ 0:00001 ð4:1þ

886 H. Tehzeeb-ul-Hassan, A. Ahmad / Electrical Power and Energy Systems 44 (2013) 880 888 Table 12 Generation and Plan of 10 Units for Proposed technique. Time Generation (MW) allocation (MW) Hour U1 U2 U3 U4 U5 U6 U7 U8 U9 U10 U1 U2 U3 U4 U5 U6 U7 U8 U9 U10 1 319 381 0 0 0 0 0 0 0 0 0 70 0 0 0 0 0 0 0 0 2 338.6 411.4 0 0 0 0 0 0 0 0 31.4 43.6 0 0 0 0 0 0 0 0 3 395 455 0 0 0 0 0 0 0 0 60 0 0 0 0 0 0 0 0 0 4 455 455 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 5 455 455 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 6 455 455 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 7 455 455 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 8 455 455 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 9 455 455 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 10 455 455 130 130 162 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 11 455 455 130 130 162 80 0 0 0 0 0 0 0 0 0 0 0 0 0 0 12 455 455 130 130 162 80 0 0 0 0 0 0 0 0 0 0 0 0 0 0 13 455 455 130 130 162 68 0 0 0 0 0 0 0 0 0 12 0 0 0 0 14 455 455 130 130 130 0 0 0 0 0 0 0 0 0 32 0 0 0 0 0 15 455 455 130 130 30 0 0 0 0 0 0 0 0 0 120 0 0 0 0 0 16 455 455 0 130 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 17 455 455 0 90 0 0 0 0 0 0 0 0 0 40 0 0 0 0 0 0 18 455 455 0 130 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 19 455 455 0 130 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 20 455 455 0 130 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 21 455 455 0 130 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 22 455 455 0 130 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 23 455 455 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 24 358.2 441.8 0 0 0 0 0 0 0 0 66.8 13.2 0 0 0 0 0 0 0 0 Table 13 Hourly total generation, reserve and profit of proposed technique. Hour k Load (MW) Generation (MW) demand allocated Profit ($) Hour k Load (MW) Generation (MW) demand allocated Profit ($) 1 18.6062 700 700 70 70 2411.3 13 23.1225 1400 1400 140 12 6186.6 2 18.6062 750 750 75 75 2601.72 14 27.4546 1300 1300 130 32 6283.6 3 18.6062 850 850 85 60 4029.29 15 23.1225 1200 1200 120 120 3857.16 4 18.6062 950 910 95 0 3713.2 16 22.0051 1050 1040 105 0 3433.04 5 18.6062 1000 910 100 0 4259.2 17 23.1225 1000 1000 100 40 3358.18 6 18.6062 1100 910 110 0 3986.2 18 22.2446 1100 1040 110 0 3173.04 7 18.6062 1150 910 115 0 3576.7 19 22.0051 1200 1040 120 0 3329.04 8 18.6062 1200 910 120 0 3258.2 20 22.0051 1400 1040 140 0 3797.04 9 18.6062 1300 910 130 0 3849.7 21 22.0051 1300 1040 130 0 4265.04 10 23.1225 1400 1332 140 0 10687.6 22 22.0051 1100 1040 110 0 4109.04 11 27.4546 1450 1412 145 0 13808.8 23 18.6062 900 900 90 10 3791.48 12 27.4546 1500 1412 150 0 16096.8 24 22.0051 800 800 80 80 3349.73 Total 121211.74 Now, sort lambda in ascending order, so that the most economical unit should be committed first, then second most economical unit and so on. Priority of the units is also required, so that for allocation of reserve most economical units must be selected instead of their simple order. It is found and saved. For each time interval (hour), each most feasible combination of units is taken by meeting minimum down time and up time constraints and then ED is done by Hamiltonian technique meeting the power limit constraints. For Hamiltonian ED a function Ed_method has been prepared separately. We call this function in main program for each most feasible combination. The complete strategy for PBUC and Hamiltonian ED is explained with the help of two flow charts. Flow chart of main program is given in Fig. 1 and flow chart of Ed_method is shown in Fig. 2. Compare this profit to the previous maximum profit. If current combination profit is more than previous maximum profit combination save the current combination otherwise retain the previous combination for maximum profit. Repeat the above for each most feasible combination, till we reach last most feasible combination. These most feasible combinations are almost equal to the number of generators. When we evaluate all most feasible combinations we have only one best combination providing maximum profit for that time interval. Then reserve power is allocated to the committed units according to their capacity and priority. Revenue, total cost and profit are calculated by meeting start up and shutdown cost constraints. Repeat the above strategy till last time interval, and find the best combination and load distribution on units providing maximum profit for each time interval. In Fig. 1 flow chart, inner loop is for combination of generators and outer loop is for time interval. Detailed Hamiltonian ED strategy is shown in Fig. 2. When we call Hamiltonian ED function we give two inputs one is the data and other is the x (status of the units). The input data contains load, generator cost curves and other necessary data required for economic dispatch. Also initialize other variables according to requirement of the function. Now check if only one unit is committed then put the load on this unit and go to the end of.the function otherwise take the first derivative of cost functions, calculate incremental cost at P max and P min of each unit. Now check more than one unit need ED? If no then set the load on that unit save the results and go to the end of the program, else find generators having maximum and minimum incremental cost from the committed units. Apply matrix solution for load distribution on committed units. Now check if load distribution on all units is within limits then save the results and end the function otherwise remove the power limit violation by considering one unit at

H. Tehzeeb-ul-Hassan, A. Ahmad / Electrical Power and Energy Systems 44 (2013) 880 888 887 Table 14 Profit comparison of proposed technique with other techniques. S. No. Other technique Profit of proposed Profit of Other technique Percentage Improvement 1 Traditional UC [7] $118869.76 $75093 58.30 2 Muller PBUC [7] $116689.76 $103296 12.97 3 LR-EP PBUC [5] $121211.74 $112818.9 7.44 4 Improved PSO PBUC [6] $121211.74 $113018.7 7.25 5 Worst fit Algorithm [9] $121211.74 $119066.3 1.8 6 Meta-heuristic PBUC [8] $120888.73 $105873.8 14.18 Table 15 Estimated execution time. Power system Hour PBUC + HED (s) 3 Units 12 0.06 10 Units 24 0.85 a time. Now again go back to the step for checking, more than one unit need ED? Repeat the above strategy till we get active power values of all the units within limits. Now save the results (active power values of all the committed units) and go to the end of the function. An active power value of each unit is our output of this function. This output is provided to the main program from where this function was being called. This technique always leads toward some final solution with small computational time, hence no stopping criterion is required for the main program. 5. Simulation results The simulation of proposed technique carried out on two simple test system adapted by different researchers [1,5,6 9]. The first system is of three units while the second system consists of ten units. Scheduling periods of 12 h is for the first system and 24 h for the second system. Unit data for each system is given in Appendix. Forecasted price, forecasted load and forecasted reserve power are also given in Appendix. Value of r is set 0.005 for three unit test system and different values are taken for ten unit test system according to the case. 5.1. Test system1: three unit 12 h test system Forecasted spot and reserve price data of three unit 12 h test system is given in Table 1 while forecasted demand and reserve power is given in Table 2. Constraints of each unit in the system are given in Table 3 and fuel cost function parameters in Table 4. For evaluation and comparison proposed technique implemented on three units 12 h test system. Allocation of power for generation and reserve to each generating unit is obtained and given in Table 5. Results obtained by conventional technique, Muller method and proposed technique are also given in tabular form. Execution/computational time comparison of proposed technique and Muller method is also presented. Table 6 contains the results obtained by traditional unit commitment. Load and reserve power demand for each and every time interval has been met by the units. Total profit obtained by this technique is $4048.8. Profit for each time interval is also given in Table 6. Results from Muller method for profit based unit commitment are given in Table 7. Total profit by this technique is $9030.5 as given in the table. This profit is almost double than the conventional technique. Computational time for this technique is also given in Table 7. The complete execution of the Muller method program took 0.078 s that is very small. This execution time can further be reduced. Proposed technique results are presented in Table 5. Total profit by the proposed technique is $8973.3 that is 2.2 times higher than the profit of traditional unit commitment technique. Table 5 also shows the computational/execution time of the proposed technique. After executing the proposed technique MATLAB program ten times, calculated the average computational/execution time. The average computational/execution time found 0.06 s. This computational/execution time for the solution of the problem is small even as compared to the Muller profit based unit commitment. Lagrange firefly PBUC takes 0.6 s for solution of three units problem for only 4 time intervals [5]. Muller method computational time is less than the LR EP PBUC and Lagrange firefly PBUC techniques, and above results show that proposed technique computational time is less than Muller method of PBUC [7]. This is because we find and save the lambda values directly where unit becomes feasible for commitment and use these values for committing the units. The graphical comparison with respect to computational/execution time of proposed technique and above mentioned techniques is shown in Fig. 3. It is clear from above discussion, tabular and graphical comparisons that with respect to computational time proposed technique is better than the LR EP PBUC, Lagrange firefly PBUC and Muller method of PBUC while it is better than traditional technique with respect to profit. 5.2. Test system2: ten units 24 h test system Proposed method is also implemented on ten units 24 h test system to solve the PBUC problem. Forecasted spot and reserve price data of the test system is given in Table 8 while forecasted demand and reserve power is given in Table 9. Constraints of each unit in the system are given in Table 10 and fuel cost function parameters in Table 11. Results achieved by traditional UC, Muller PBUC, LR-EP PBUC, improved PSO PBUC, worst fit algorithm, Metaheuristic PBUC and proposed technique are presented. Profit comparison by IEEE 10 units 24 h test system, of proposed technique with traditional and other techniques is described with the help of tables and graphical data. Execution/computational time comparison of proposed technique with other techniques is also exposed for IEEE 10 units 24 h test system. The complete power generation plan and allocated reserve of each unit for 24 h by proposed technique is shown in Table 12 for all the time intervals (hours). Table 13 contains lambda, load demand, total generation, reserve demand and total reserve allocation and profit for each time interval (hour) calculated by proposed technique. Profit obtained by different techniques and proposed technique for 10 units 24 h test system is presented in Table 14. In some cases reserve, spot price and reserve price conditions may be different from conditions given in Table 9 but the profit of proposed technique and other technique at each row of Table 14 is with same conditions. This is why each technique profit is compared independently. Detailed results and conditions can be viewed by the reference mentioned with each technique in Table 14. Profit obtained

888 H. Tehzeeb-ul-Hassan, A. Ahmad / Electrical Power and Energy Systems 44 (2013) 880 888 by techniques mentioned at S. No. 1 of Table 14 is with zero reserve condition for all the time intervals. It is also clear that proposed technique improved the profit by 58.30% than the traditional technique. The graphical comparison of total profit by proposed and traditional technique is also presented in Fig. 4. Results at S. No. 2 are with the condition of defined hot and cold start up cost of each unit and zero reserve requirement for all the time periods. With these conditions improvement in profit by proposed method is 12.97% than the Muller PBUC method. The comparison of both techniques with respect to profit is shown in Fig. 4 graphically. For the load and reserve conditions mentioned in Table 9, total profit achieved by techniques mentioned at S. No. 3, 4 and 5 is presented for 10 unit system. The improvement of profit by proposed technique came out 7.44%, 7.25% and 1.8% as compared to LR-EP PBUC, Improved PSO PBUC and Worst fit Algorithm techniques respectively. This comparison is also shown graphically in Fig. 4. At S. No. 6 of the table, profit by Meta-heuristic PBUC and proposed technique presented with 80% spinning reserve mentioned in Table 9. Improvement of profit for this problem by proposed technique as compared to Meta-heuristic PBUC for 10 unit system observed 14.18%. Literature shows that LR PBUC takes number of iterations for their final solution thus its computational time is large and sometimes suffers convergence problem. Meta heuristic technique takes 131 s for the solution of the same problem [8]. Proposed technique solves the problem in less than one second (most of the time 0.8 0.9 s) as shown in Table 15. 6. Conclusion This paper presented the solution of UC and ED problem. PBUC and Hamiltonian Economic Dispatch (HED) have been implemented for the solution of unit commitment and economic dispatch of IPPs. The proposed technique is tested on IEEE 3 unit 12 h and 10 unit 24 h systems. Different techniques have been used in past for the solution of unit commitment and economic dispatch problem. In this paper, for getting maximum profit for each time interval PBUC is used and Hamiltonian technique is used first time for economic dispatch. The obtained results by using PBUC and HED are better than some previous results reported in research papers both for execution time and profit. This reflects that the implementation of this work is better than most of the previous implementations. This implementation is favorable specially, where cost curves of generating units are close to each other or sudden changes in load and reserve demand. In proposed technique, analysis of most feasible combinations of the generating units with respect to profit carried out including HED. This technique always produces the solution of the problem while LR PBUC may face convergence problem in some cases. Execution time of PBUC and HED is less than most of the PBUC techniques mentioned in literature. Profit of the proposed technique is comparable to the traditional technique. Results conclude that the proposed technique is more attractive for IPPs with respect to the profit as compared to traditional techniques and with respect to execution time as compared to some advance techniques as well. Results also show that in order to find the near global solution for UC and ED as an objective function, PBUC HED can be counted among some efficient techniques. 7. Future work For future, proposed technique instead of traditional UC and ED can be implemented for planning with respect to profit before installation of generating units for IPPs. Some other constraints according to the environment, unit characteristics, market trends and power system of the country can also be included for more realistic and practical analysis. For online application, PBUC HED can be used because its computational time is small and can be improved by different methods or programming it in some other language i.e. assembly language. For large power systems where computational time becomes of more significance proposed technique can be utilized. Moreover, computational time of proposed technique can further be reduced by using the already done analysis during the solution of the problem. In this way, it can be used for very large power systems as well. Appendix A See Tables 1 15. References [1] Attaviriyanupap P, Kita H, Tanaka E, Hasegawa J. A new profit based UC considering power and reserve generating. IEEE; 2002. [2] Allen E, Ilic M. Price-based commitment decisions in the electricity market. London: Springer; 1999. [3] Hassan HT, Imran K, Akbar MS, Ahmad I, Siddiqi YM, Impacts of proposed utilization of various indigenous coal reserves for power generation on production cost and environment. Proceeding of institution mechanical engineers, Part A. J Pow Energy 2010;1(1). [4] Contreras-Hernández EJ, Cedeño-Maldonado JR. A sequential evolutionary programming approach to profit based Unit commitment. In: IEEE PES transmission and distribution conference and exposition Latin America, Venezuela; 2006. [5] Pathom Attaviriyanupap, Hiroyuki Kita, Eiichi Tanaka, Jun Hasegawa. A hybrid LR EP for solving new profit-based UC problem under competitive environment. IEEE Trans Pow Syst 2003;18(1). [6] Yuan Xiaohui, Yuan Yanbin, Wang Cheng and Zhang Xiaopan. An improved PSO approach for profit-based unit commitment in electricity market. IEEE/PES transmission and distribution conference & exhibition: Asia and Pacific Dalian, China; 2005. [7] Chandram K, Subrahmanyam N, Sydulu M. New approach with Muller method for Profit Based Unit, Commitment, ÓIEEE; 2008. [8] Mori H, Okawa K. A New Meta-heuristic Method for Profit-Based Unit Commitment under Competitive Environment. [978-1-4244-2235-7/09/ $25.00 Ó2010 IEEE]. [9] Amudha A, Christober Asir Rajan C. Effect of reserve in profit based unit commitment using worst fit algorithm. [978-1-61284-764-1/11/$26.00 Ó2011 IEEE]. [10] Allen J. Wood, Bruce F. Wollenberg. Power generation operation and control. 2nd edition. Jhon Wiley & Sons Inc.; 1996. [11] Carpentier J. Optimal power flow. Electr Pow Energy Syst 1979;1(1):3 36.